Vijay Gill — one of the brains that oversees Google's epic internal network — has questioned the economics of so-called cloud computing. Or least, the sort of cloud computing practiced by Amazon.com, whose EC2 service offers up instant access to compute power via the interwebs. If your infrastructure is in use around the clock, …
Anther savings is that if you're using Amazon you're stuck with a few different memory sizes of virts that you can get from EC2: 1.7GB, 7.5GB, 15GB, etc. If you need a bunch of 4GB virts then you're stuck buying 7.5GB RAM servers and trying to politically deal with bundling up apps on the same O/S instance, or else just wasting RAM.
You'll get better "compression" by being able to size the virts yourself by running your own internal cloud.
I've also been doing these calculations repeatedly because our board of directors seems to be infected with irrational lust for the cloud, and keep on determining that break-even on a cash-flow basis comes after about 3-4 months for 100% duty cycle instances -- so Vijay is probably being conservative at what can be achieved in the real world.
However, if your IT department spends millions on expensive enterprise-class tools and money is flying out the door to VMware, EMC, NetQOS and every other IT vendor out there, then it may be cheaper to go with the cloud -- but in going with the cloud you are getting Amazon's framework build on top of Xen and open source or in-house tools. You are not getting infrastructure engineered to "five 9's" you are getting infrastructure engineered to "you bet it'll fail -- so spin up another one!"
If you run the numbers on cloud and find that it makes economic sense, you should fire your IT staff and start over from scratch with people who understand cheap open source.
Welcome to 1970.
You can either pay a bureau and they own the overheads, shared across multiple customers not just you, but you still carry the risk and the responsibility when it doesn't work as intended.
Or you can put a PC on every desk, and pay way over the odds for the compute power and storage which sits idle for 99% of the time, and pretty software which is defective by design.
Or you can take control, do most of it in house, as far as possible with free software, and know exactly what you're getting for your money and maybe be able to fix it when it misbehaves.
Stuck in a time loop...
Using "reserved instances" totally changes things
As you say: "the Google man uses Amazon's standard pricing for instances, rather than its reserved pricing."
There's a *vast* difference between those prices.
$0.68 per hour without reserving an instance,
$0.24 per hour when you reserve an instance (for a $2800 once-every-three-years reservation fee).
This knocks the monthly price down from $118k to under $55k.
...I can't disagree with Mr. Gill more.
Modern data-centers use elastic computing and dynamic machine provisioning already, either on premise or on the public cloud. The simulation Mr. Gill is proposing is dense nonsense.
Of course this is about bashing AWS IaaS approach and promoting Google PaaS offering.
How much of one's reputation can be thrown to pigs and dogs to support marketing strategy?
No, I don't want to know, actually...
Paris because she has a reputation.
Is this really a surprise?
Anyone who's run the numbers on on-demand computing vs rolling your own knows that it's much cheaper to roll your own. The only place where on demand is cheaper is for bursted capacity... Yeah, people argue about staff overhead, but it's not like running on cloud services frees you to have no staff.
I guess the real problem is that people don't actually calculate the cost of things....
i think it makes sense...
economically that amazon will price things so what they charge pays the bills *and* turns them a profit. So, they have large economies of scale, but at some point it must be cheaper to keep everything in house, or at least keep most in house and rent ec2 time for demand spikes.
Way over my head...
...I can't imagine these sorts of server setups, never mind know enough to think of maintaining them, however it seems to me that this comparison sits on a raw cost for bandwidth heavy services while glossing over a few small things - reliability (what's the uptime?), security (unauthorised people can't access the data, and if they do damage will be minimal), and integrity (are full backups taken? how often? how many are archived? are these backups actually tested?).
Statement of the bleeding obvious
Of course Amazon does not make financial sense if you run it flat out. Same numbers for 10 or 1% utilisation are hugely in Amazon's favour. 10% is a dream come true for a small or medium business e-commerce front-end. 1% is more like it.
From that perspective Amazon is actually very fit for service.
Kind of makes sense
If you are using the machines 100%/24/7, your costs of running them are going to be pretty much the same as Amazon's costs + Amazon needs to make a profit out of it, and Amazon prices on the assumption that they aren't going to get 100% utilisation of the machine - they need to have spare machines available at a moments notice for anyone who wants them. The whole point of the set up is that you can share the cost of running the machine with other people who might want it when you don't. If you are not sharing, there is no point in going to Amazon.
But this duty cycle goes to 111%
What happens when you need *more* cycles that you have available in-house? When you get Dugg or Slashdotted, serving 404s is throwing away your one chance to cash in.
As an EC2 user...
I work with a small software company which uses EC2. Of course when you look at the numbers for hardware and bandwidth its much cheaper to have hardware in-house.
But hardware and bandwidth are the cheap elements. Much more expensive is the cost of staff to maintain that hardware/infrastructure. If you're part of a bigger company or if you already have the expertise needed to maintain the infrastructure, ensure your ISP always provides adequate bandwidth at edge locations, have the reserve capacity in the event of failure, the storage available to backup, etc. then Amazon might not be cost effective. For us, even without taking into account the reduced pricing of reserved instances the question is never EC2 or in-house its EC2 or some other equivalent service.
Even if a company does have the required expertise in-house its going to be worth re-considering the economics. Economists tell businesses they have to focus on their core competency and out-source other tasks where possible even if the business would be better at that out-sourced task because it means the business can focus on doing even more of what they are expert at. This is especially true for small companies like ours where there's a tendancy to do everything. In our case running a data center is non-core so we out-source it to Amazon.
Of course the economists logic is fuzzy at the edges. We have to have some in-house capability. But we are limiting that to a Windows 2008 domain server (and backup), test rigs and staff machine maintenance. However our core compentence requires we understand Microsoft's server and desktop OSs.
a lot more to factor
Going on just the information in the article, did they factor in all the indirect overheads like insurance on the equipment, physical security, desks and all that sort of crap.
Most businesses want to run a business and not a data centre, getting into the whole do it yourself IT thing tends to distract the organisation from their core focus.
tied to resources
the calculations also split the cost of the physical kit over a 3 year cycle... that ties you to capital expenditure and maintainance costs for those 3 years...by the 3rd year you're replacing the kit? thats then an ongoing issue...as all of us with large data centres already know.
no, the Amazon model is compelling - but, as pointed out elsewhere in this thread - can be done cheaper in-house and with hosting partnerships.
I believe one factor in cost that isn't consider is the cost of the cloud sofware license to support your own cloud computing compared to Amazon. Can anyone elaborate on the cost of VMWare vsphere and its tools?
EC2 isn't as great as it sounds
Personally - I think EC2 hype is blown WAY out of proportion. I read all the time that people flock to EC2 like lemmings thinking it is the most cost effective cloud platform out there. I happen to know from experience it certainly isn't the most friendly to use. To the "Google Man's" point, whether you agree with him or not, EC2 isn't the end all be all of cloud platforms. I found this 30,000 foot comparison matrix of the largest cloud platform providers that shows there are many things to consider when factoring cost and eventually adoption to the cloud. Just sayin'....
Duff numbers, fatally flawed assumptions
So, his cost figures are completely wrong (using the per-hour pricing for EC2 against long-term commitment pricing for the colo option), his assumption of 100% usage for any Internet situation is nothing short of absurd when even mainframe shops with pre-planned predictable batch workloads don't come close, and he's missed out the whole "elastic" bit of the elastic compute cloud. Apart from that, nice try.
For the real world, on the other hand, instead of paying for, say, ten colocated servers and bandwidth constantly and getting hammered by the odd traffic spike, I can pay less money per server to have six servers handling routine traffic - and crank up to 20 servers at a minute's notice for the busiest couple of hours, so I'm providing a much better service to customers for less money.
What next, Gilette blogging that it's worse for you to drop a mains-powered electric shaver in your bath than a manual blade razor!?
- Product Round-up Smartwatch face off: Pebble, MetaWatch and new hi-tech timepieces
- Geek's Guide to Britain BT Tower is just a relic? Wrong: It relays 18,000hrs of telly daily
- Geek's Guide to Britain The bunker at the end of the world - in Essex
- Review: Sony Xperia SP
- Dell's PC-on-a-stick landing in July: report